def test_simplex_2b(self):
# Now check for LEAST enjoyable diet.
logging.info("\nFINDING VERTEX FOR 2(b) BY FLIPPING SIGN ON c")
self.c *= -1.0
x = undertest.all_inequality(self.c, self.x_0, self.constraints)
# Now check for optimality.
original_constraint_idxs = range(4)
self.assertTrue(np.all(
np.dot(self.A, x) >= _build_ndarray(self.b, original_constraint_idxs)))
logging.info("Checking vertex for 2(b). Ax = {0}, b = {1}".format(
np.dot(self.A, x), self.b))
def test_simplex_2a(self):
"""
Use the initial vertex x_0 to run simplex.
"""
logging.info("\nFINDING VERTEX FOR 2(a)")
x = undertest.all_inequality(self.c, self.x_0, self.constraints)
# Now check for optimality. x should satisfy constraints and yield non-negative lambda.
original_constraint_idxs = range(4)
self.assertTrue(np.all(
np.dot(self.A, x) >= _build_ndarray(self.b, original_constraint_idxs)))
logging.info("Checking vertex for 2(a). Ax = {0}, b = {1}".format(
np.dot(self.A, x), self.b))
def test_simplex_toy_problem(self):
"""
The notes include a toy problem. Test the simplex method on this problem, make sure it
returns the right solution.
"""
raw_constraints = [[0, -1], [1, 2], [-1, 1], [1, -1]]
b = [-1, 0, -1, -1]
working_set = [0, 2]
constraints = undertest.Constraints(raw_constraints, b, working_set)
c = np.array([[1, -1/3]]).T
x_0 = np.array([2, 1])
# Check the optimal solution
x = undertest.all_inequality(c, x_0, constraints)
nptst.assert_array_almost_equal(x, np.array([-2/3, 1/3]))
